From: Heiko Carstens Date: Mon, 14 Jul 2008 07:57:28 +0000 (+0200) Subject: [S390] sclp: fix possible deadlock on cpu rescan. X-Git-Tag: v2.6.27-rc1~1112^2~65 X-Git-Url: http://pilppa.com/gitweb/?a=commitdiff_plain;h=b9732ca1cb2c18129d2f984b8de02db141b7b830;p=linux-2.6-omap-h63xx.git [S390] sclp: fix possible deadlock on cpu rescan. smp_rescan_cpus() calls get_online_cpus() from a multithreaded workqueue context. This may deadlock. This is the same bug as in arch/s390/kernel/topology.c. This patch can be reverted as soon as Oleg's patch gets merged. Signed-off-by: Heiko Carstens Signed-off-by: Martin Schwidefsky --- diff --git a/drivers/s390/char/sclp_config.c b/drivers/s390/char/sclp_config.c index ad05a87bc48..7c9b39cee59 100644 --- a/drivers/s390/char/sclp_config.c +++ b/drivers/s390/char/sclp_config.c @@ -8,6 +8,7 @@ #include #include #include +#include #include #include #include @@ -40,9 +41,19 @@ static void sclp_cpu_capability_notify(struct work_struct *work) put_online_cpus(); } -static void __ref sclp_cpu_change_notify(struct work_struct *work) +static int sclp_cpu_kthread(void *data) { smp_rescan_cpus(); + return 0; +} + +static void __ref sclp_cpu_change_notify(struct work_struct *work) +{ + /* Can't call smp_rescan_cpus() from workqueue context since it may + * deadlock in case of cpu hotplug. So we have to create a kernel + * thread in order to call it. + */ + kthread_run(sclp_cpu_kthread, NULL, "cpu_rescan"); } static void sclp_conf_receiver_fn(struct evbuf_header *evbuf)